Automated Container Probability Project 2020+

[DaringDeer 656]

Thanks to @ILiveOnTheChatBox123 for restarting the manual collation of container data in his thread and @Turrets for curating responses.

I wanted to see if I could create a Google Form where players could easily enter the results of their container opening and results are automatically fed into a Google Sheet to show results and probabilities.  After several iterations, i am ready to share it.  The entry form allows entry of up to 15 containers at a time, irrespective of how many you are opening at once.

If you have any ideas for improvements or other visualizations of the data let me know.

Form:  https://docs.google.com/forms/d/e/1FAIpQLSe2ZG3EblCt1fCCrhyDVXNhALMVQfUp4DnsJb3dSosY5TmKRg/viewform

Results:  https://docs.google.com/spreadsheets/d/1h9qZ9SzO2QWGa0TMIGZPdZ6bHxia9NRGbINPuaJRBJc/edit#gid=0

Form:

Spoiler

 

Edited May 13, 2020 by DaringDeer
Recognized ILiveOnTheChatBox123

[DaringDeer 656]

Live Results

Spoiler

Let me know if these images stop working.  They are automatically taken screenshots from results screen to include here.

 

Edited June 4, 2020 by DaringDeer
Added Exotic - Drone option. Updated live links. Changed to bar chart.

[mjmj5558 2,346]

It would be good to see the overall chance of common, uncommon, rare etc.

[ILiveOnTheChatBox123 936]

2 hours ago, DaringDeer said:

Thanks to @Turrets for restarting the manual collation of container data in his thread.

 

Old topic: (see who really started the manual collection)

Once my topic is closed, I hope your topic works out better than mine.

Wouldn't it be better to also show the total rate for each category, meaning all the items in that category added up together?

Edited May 12, 2020 by ILiveOnTheChatBox123

[coconuttree 519]

Seperate probability for each item is better. Since apparently they are not the same within categories.( And some items are useless to me)

[Turrets 6]

So... what about fake entries? I feel that an actual post can be trusted more than an anonymous entry.

[CaptainKiller 67]

10 minutes ago, Turrets said:

So... what about fake entries? I feel that an actual post can be trusted more than an anonymous entry.

Someone opening the containers starting from recruit and making independent results may be a better way to gather results......

 

[DaringDeer 656]

As far as dealing with fake entries, I can view the raw captured results and remove any that are suspicious.  If no one is opposed to my sharing it, I can publish the raw data in read-only format for anyone to see.  It includes the tanki player's name.

If anyone else is interested in helping to curate the data, I can give them edit privileges to the Google Sheet.  PM me your Google ID.

In 2nd post I did add semi-live image links (cached for 3 hours) to screenshots of the results.  I'm not sure how reliable the screenshot links will be.

[coconuttree 519]

Btw, in the past, the test server has given a million tankions. If it happens again, this can be used to get massive amounts of containers and add to the database.

(However there is a chance that the drop rates will be different in the test server) 

[mjmj5558 2,346]

4 hours ago, coconuttree said:

Btw, in the past, the test server has given a million tankions. If it happens again, this can be used to get massive amounts of containers and add to the database.

(However there is a chance that the drop rates will be different in the test server) 

Yeah, so we should keep this topic only normal server.

[Michael828 816]

Can the overall probability chart be a bar chart instead? Because I feel it's clearer than a point / scatter one.

Also requesting screenshots as a method of proof is fairly reliable and can't easily be faked (that is of course if they include the date and their usernames in them).

 

Good luck!

[BigCar 54]

free containers time 

[printing_now 59]

well... no containers here. I'll see what i can do with it later.

[coconuttree 519]

anybody notice an increased probability of getting rarer tiers or is it just me?

[TheCongoSpider 8,942]

1 hour ago, coconuttree said:

anybody notice an increased probability of getting rarer tiers or is it just me?

You guys are getting rare items?

[coconuttree 519]

16 minutes ago, TheCongoSpider said:

You guys are getting rare items?

usually no. but in the past 3 days much more than usual

[mjmj5558 2,346]

7 hours ago, coconuttree said:

usually no. but in the past 3 days much more than usual

You are lucky.

[wolverine848 13,097]

9 hours ago, TheCongoSpider said:

You guys are getting rare items?

In 30+ containers the best I've gotten was 3 days Premium and a shot-colour (don't have the turret of course) 

[DaringDeer 656]

My anecdotal experience is that recent openings have been better than older ones.  I opened 20 containers yesterday and received 2 epic hull alterations.  A couple days ago I opened the free containers on a few low accounts (8 containers over 4 accounts) and received an epic hull alteration.  Certainly in the last few days I am seeing many more people wearing hull alterations in Legend battles.

I think the numbers are going to fluctuate a good amount for the rarer (<2%) rewards until we get more data.  Pretty consistently seeing ~65% are common and ~25% are uncommon.  That leaves just 10% of getting something interesting. 

At some point in future I can create views for different periods of time. 

[kydapoot 570]

This is just my opinion, but I think the google form should only accept opening containers 15 at a time. People won't be as consistent reporting one or two containers at a time, and so there will be some unintentional bias that arises from the tendency to report interesting finds and blow off common ones. Bundling the openings would reduce the variation in rarity proportions.

Edit: The form overall is more convenient than making a forum post, so maybe this is a moot point now. What do you think?

Edited May 25, 2020 by kydapoot
Note about form

[kydapoot 570]

Also, I have a question about the statistical valididy of collecting percentages for events with changing probabilities. Every time you get a paint, augment, shot effect, or skin, your chances of getting that a similar item in the future presumably go down (for example, now that I have all of the football paints, I have a 0% chance of getting one from the rare tier). Is there any information about the changing behavior of containers as certain rewards are all obtained?

It's possible that the probability of getting a rare paint, for example, stays the same until they are all gone, and then that reward is replaced either with another rare reward, an uncommon reward, an epic reward, or another random draw altogether. It's also possible that the probability of getting a rare paint (or epic/legendary, skin, augment, effect) is dependent on the number you have yet to obtain; like you're pulling them from a pool. Which do you think is more likely)

Unless we know this, do we just have to accept the results of this data as reflective of what the average tanker will get, and understand that people with a lot of (or all of) the paints, skins, effects, or augments could have moderately different results?

[DaringDeer 656]

The results are only going to be as good as the data people enter.  I think because the results are anonymous it isn't likely that people will only report good or bad openings.  They're more likely to post to the forum, tell a friend, etc..  Unfortunately, not a lot of people are using the form.  (Please spread the word.)  There have been less than 2k total openings entered and probably 1/3 are from me (either mine, a friends, or from a youtube video).  Still a relatively small sample.

The account opening the containers has to have an effect on the results but it would take a lot of data to determine that.  On this account I have virtually all the equipment except hull alterations.  I've gotten many more hull alterations on it than I have on my lower account (jslifer).  On my lower account, I actually get better containers on average, it's just that those can also be paints, shot effects, turret alterations, etc.. 

My guess based on limited data below is that every possible reward has a predefined chance to occur.  Because DaringDeer already owns more Rare/Epic/Legendary/Exotic items than jslifer, the likelihood of getting those types of containers is lower for DaringDeer.

 

DaringDeer:

Common	160	61.78%
Uncommon	68	26.25%
Rare	19	7.34%
Epic	10	3.86%
Legendary	2	0.77%
Exotic	0	0.00%

 

jslifer:

Common	224	63.64%
Uncommon	69	19.60%
Rare	34	9.66%
Epic	21	5.97%
Legendary	3	0.85%
Exotic	1	0.28%

[DaringDeer 656]

Some other data points of unnamed players (all Legends) with 150+ recorded openings.  The last guys has been very unlucky.

Common	138	62.73%
Uncommon	62	28.18%
Rare	16	7.27%
Epic	3	1.36%
Legendary	1	0.45%
Exotic	0	0.00%

 

Common	108	66.26%
Uncommon	30	18.40%
Rare	13	7.98%
Epic	8	4.91%
Legendary	1	0.61%
Exotic	3	1.84%

 

Common	152	72.38%
Uncommon	41	19.52%
Rare	15	7.14%
Epic	2	0.95%
Legendary	0	0.00%
Exotic	0	0.00%

[mjmj5558 2,346]

4 hours ago, DaringDeer said:

Some other data points of unnamed players (all Legends) with 150+ recorded openings.  The last guys has been very unlucky.

Common	138	62.73%
Uncommon	62	28.18%
Rare	16	7.27%
Epic	3	1.36%
Legendary	1	0.45%
Exotic	0	0.00%

 

Common	108	66.26%
Uncommon	30	18.40%
Rare	13	7.98%
Epic	8	4.91%
Legendary	1	0.61%
Exotic	3	1.84%

 

Common	152	72.38%
Uncommon	41	19.52%
Rare	15	7.14%
Epic	2	0.95%
Legendary	0	0.00%
Exotic	0	0.00%

I am like the last guy: i never opened an exotic or legendary.

[Blutaar 1,904]

On 5/26/2020 at 10:03 AM, mjmj5558 said:

I am like the last guy: i never opened an exotic or legendary.

I would rather say the second guy was extreme lucky. Even on the test server I bought around 1500 containers and only managed to get 1 exotic container